This is a patient-oriented career development proposal designed to provide the candidate with advanced training, protected research time, and mentored research experience. The research aims focus on Borderline Personality Disorder (BPD), which is a serious and debilitating mental illness that affects 2-6% of the population and increases risk of suicide. Current treatments have limited efficacy and there are no FDA- approved medications for BPD. The candidate's long-term career goal is to improve the treatment of BPD and other disorders with significant social symptoms by using computational psychiatry to better predict prognosis and best treatment match. To continue her progress toward this goal, the candidate proposes a detailed plan for training, including expert mentorship and three supervised research aims. The training plan is designed for the candidate to gain expertise in neuroimaging, computational modelling, and learning theory. The candidate will also build on her strong foundation in clinical research and treatment of people with Borderline Personality Disorder and in statistical approaches to data analysis. The mentorship team has extensive expertise in clinical psychiatry research, fMRI study design and analysis, and computational modelling of learning. They will provide the candidate with the resources and supervision needed to advance toward her research goals. These studies leverage recent advances in computational psychiatry. Through three hypothesis-driven research aims, subject behavior and brain activation will be tested during an interactive social learning task. These aims test learning under volatility (when the environment is unpredictable) in BPD compared to BPD+PTSD, PTSD, and trauma-exposed healthy controls. Aim 1 tests the role of anterior cingulate cortex for signaling volatility, and interaction between amygdala and anterior cingulate in this setting. Aim 2 tests learning patterns in BPD versus comparator subjects to identify illness-specific phenotypes. Aim 3 tests how neural and behavioral markers of learning relate to social functioning. Next steps will be to refine a computational model of learning in BPD to predict prognosis, predict best treatment match for an individual, and test novel biological treatment targets. Improving our mechanistic understanding of interpersonal symptoms will improve our clinical treatments and significantly reduce suffering for millions of people with BPD and their families.